When I press the right "window" key, between AltGr and the menu key on my
Swedish keyboard, it generates mouse button events rather than key events.
The precise button events are different on different hardware. I attach a log
where it generates press and release of button 6 when I press the key, and press
and release of button 7 when I release it. When I tried on different hardware,
it generated press an release of button 4 only.
This is with Fedora Core 4 test release 2. Package version is xorg-x11-6.8.2-19.
I am not completely sure it is an X bug, and not in some underlying system. But
it does not seem to be hardware. It happens on two differnt setups (both with a
Swedish keyboard layout). With the same hardware the key works correctly in
other environments (older Fedora, Knoppix, Windows).
Created attachment 2445 [details]
The configuration file on one of the systems
Created attachment 2447 [details]
Trace when running a fresh X server with only xev
I started the X server with xev as the client, moved the cursor to the xev
window, typed the keys at the bottom of the keyboard in sequence (left control,
left window, Alt, space, AltGr, right window, menu, right control). Then I
switched back to the original server (Alt-Ctrl-F5) and killed the xev.
I see this too, on current rawhide
(In reply to comment #3)
> I see this too, on current rawhide
Still present on the released FC4, updated to xorg-x11-6.8.2-37. Took me a
while to debug why I had lost my compose (multi_key) capabilities, which I put
on that key.
As can be seen in the discussion in the corresponding report in Red Hat's
bugzilla, there are more and more evidence this might be a bug in the FC4 kernel
rather than in Xorg. See the URL for further details.
I am experiencing the problem on a Gentoo system.
The versions I'm running are xorg-x11-6.8.2-r2 (6.8.2 with a bunch of patches)
and vanilla-sources-2.6.12_rc1 (unpatched 2.6.12-rc1). I will upgrade to the
latest kernel to see if it makes a difference.
I just upgraded to 2.6.13-rc1 and the problem has disappeared.
With an updated kernel 2.6.12 release for Fedora, the problems went away for me
too. Together with the other evidence here, I think it is safe to assume it was
a kernel bug all the time.